SoC% false accuracy? State of Charge

Announcement - Help us fight the BOTS! Please report all spam including stuff in your inbox!

gauss163

Member
Joined
Jun 28, 2020
Messages
278
completelycharged said:
Issue with typical joule / coulomb approach is it's your battery cycle efficiency that limits the accuracy unless you reset the counter each charge cycle [...]

Li-ion has very high coulombic efficiency (typically over 99.5%) so that's not really the worst problem with coulomb counting fuel gauges. Rather, it's correcting for losses not measurable by the gauge, e.g. self-discharge (esp. if it's idle for long periods) and also losses due to loads too small to be measured by the gauge (e.g. BMS quiescent current), which may require a full learning cycle to recalibrate (in the simplest gauges), and such full cycles may occur only rarely (which can be alleviated to some degree by using various methods of voltage modelling). See this Barsukov bookfor much more on fuel gauges.
 

daromer

Moderator
Joined
Oct 8, 2016
Messages
5,659
Yes but i would rather do it when fullt charged ir 100%
 

Redpacket

Member
Joined
Feb 28, 2018
Messages
1,309
Agree the most practical method to keep the SoC meter close is for it to reset at full charge like daromer says.
The problem of knowing when your battery bank dropped capacity a bit from use/aging, etc has to be found with a full cycle & adjusting the coulomb counters AHr value.
 

OffGridInTheCity

Active member
Joined
Dec 15, 2018
Messages
1,746
Honestly - I don't pay any attention to SOC but instead, focus oninverter on/off voltages and the power/voltage-range produced each day. The current batteryvoltage is perfectly adequate (for me) to have a spit-ball notion of where things are in the charge/discharge curve.
 

daromer

Moderator
Joined
Oct 8, 2016
Messages
5,659
But the voltage during load is hard to counter how much you actually have left. Proper SOC is alot easier :) Especially on LiFe where its impossible to read charge-level based on voltage during run-time

I never look at voltage during normal behaviour. Only when doing a full run below 10% soc because thats where my system start to drift.
 

OffGridInTheCity

Active member
Joined
Dec 15, 2018
Messages
1,746
daromer said:
But the voltage during load is hard to counter how much you actually have left.
Ah - but remember that I'm nearing2 years of detailed monitoring and things are stable - so voltage is a perfectly accurate 'enough' predictor for me. One of the things I'm tracking is ah/v - hoping that this will be an early indicator of battery bank degradation.

But yes, when you don't have a long history or stable system or are just interested in SOC- I can see where its an interesting metric.
 

daromer

Moderator
Joined
Oct 8, 2016
Messages
5,659
Yes but the voltage differ due to current. You need to have an equation based on current/volt to be able to do it but yes on LiIon that you test you can basically do it but its not as accurate as Colloumb counter is.
 

Redpacket

Member
Joined
Feb 28, 2018
Messages
1,309
So it sounds like OffGrid has a "lesser tech" but functioning method, a sort of lookup table requiring a detailed understanding of your batteries & load currents, etc.
If that works for you, great!
The coulomb counter methods are more hands off & automatic + sound easier to use (mine is!)
Whatever works, but know your SoC!
 

OffGridInTheCity

Active member
Joined
Dec 15, 2018
Messages
1,746
Yea - I think I'm missing the focus on the usefulness of SOC - honestly interested in understanding more about this if you'll try to help me. Maybe I am but I don't 'name it' that way?:)

My PV is 12.85kw andbattery bank is 80kwh.To operate, I twiddle with the charge/discharge voltage range to 1) consume all PV power produced and 2) to extend battery life. Right now -batterycut-off is49.5v (3.54v/cell)and float is56.6v(4.04v). On average, I only reach 3.90v/cell as the daytime consumption is high enough I don't hit 4.0v/cell max very often.

This cycle repeats each day - and I track stats daily such as total PV power produced, total run time of inverters, andkwh(s) of output from the inverters, ah/v, and some others. This gives me daily, monthly, yearly stats such as:

For the year so far....
* Have produced 13,023 kwh of PV
* Have consumed 10,741 kwh of power from the inverters.
* This is82.5% efficiency ( inverter-kwh / pv-kwh ) **Needs tweaking to reach 85% which is doable
* Average hi battery voltage = 3.90 (low is always 49.5v)
* Average DOD is ~ 46% ... **This is anSOC affiliated number**

I'm interested in DOD but SOC as a numberdoesn't figure into daily operational focus... What I focus on is max pack voltage differences (are they healthy),low/hi battery voltage range for 100% PV consumption and long battery life.

What am I missing about SOC focus?
 

daromer

Moderator
Joined
Oct 8, 2016
Messages
5,659
SOC tells you the factual capacity in a system. Yes you can use lookup table/formula to calculate it from voltage but most people use colloumb counter where you count the energy going in and out of something. This is often used with a shunt.
Why? Because it always know the state

Doing this based on voltage on lithium or any battery requires either:
* To have 100s of tables or 1 very very advanced formula.
* Or just doing it during resting.

Its the same principle done on LA or RC batteries :)

The thing is that under load the voltage drops and then presumable show you a lower SOC value if you translate it from voltage than when resting. Same goes with charging. You can basically be at your max voltage but the battery isnt full because you are still pumping in xx current :)

Example my battery bank can be at 4.0V per cell but my SOC tells me its at 92%. That shows me that i still have xx kWh left to charge ;)

So to translate voltage to SOC the easiest is resting. But this does not work on LiFe. LiFe stays at its middle voltage over almost 80% of its charge/disharge curve :) And varies with load and temperature. So on those its even harder to do it with voltage. You get a rough estimation in the ends only.

What you are missing about SOC i dont know. Not everyone care about it but i do. Then i know if my battery is full. I can also easily calculate how much left is to charge. Or during the night if the rest of the charge will last. Based on lets say if i got 20% left than i know i have 35kwh left to use. Easy to calculate ;)
 

gauss163

Member
Joined
Jun 28, 2020
Messages
278
OffGridInTheCity said:
What am I missing about SOC focus?

The mapping from (rest) voltage to SOC depends on the chemistry, e.g. for Sanyo LiCO 3.54V is 1% but it is 28% for Pana NRCB's. Similarly 3.90V is 28% vs. 75%. So when we are discussing matters that really depend on SOC (e.g. balance your cycles around 50% SOC) it is best to use SOC numbers since they apply generally - independent of chemistry. And for Frankenstein packs with a motley mix of chemistries the V->SOC mapping is not even known a priori.

daromer said:
SOC tells you the factual capacity in a system. Yes you can use lookup table/formula to calculate it from voltage but most people use colloumb counter where you count the energy going in and out of something. This is often used with a shunt.
Why? Because it always know the state [...]

No, simple coulumb counting does not always know SOCsince e.g. it cannot see self-discharge losses (and also currents below its resolution, such as BMS quiescent current).

To learn how battery fuel gauges work I highly recommend this Barsukov bookby a battery guru at TI = Texas Instruments. He is the designer of TI's impedance tracking fuel gauge algorithm - by far the most widely used algorithm in laptops. The chapter on fuel gauges there is one of the most accessible introductions to the subject.
 

OffGridInTheCity

Active member
Joined
Dec 15, 2018
Messages
1,746
I see (better). One reason Ishy away from SOC in the Batrium context is that it doesn't seem to maintain any accuracy. On the Shunt page -> Metrics... I can set an explicit "State of Charge" - e..g 75% right now. But that doesn't seem to stay accurate over time.

image_kxhtrr.jpg


The Re-Calibrate in Bypass is no good because I don't run balance / never get to the top where balance would kick in.

The Re-Calibrate Low SoC / Hi SoC - I have it ON, butI don't know what to set it to to make it retain accuracy - I don't get it.:( Right now I just have it 'out of the way' at -2.5% and 102.5%.

Maybe you could explain how to set these to keep the SoC accurate? or maybe there's something else to set?

image_goqzwc.jpg



*The Nominal Capacity of 1560Ah is correct.
 

daromer

Moderator
Joined
Oct 8, 2016
Messages
5,659
the SOC need to be and should be recallibrated either when its fully charged or empty. For instance when i have reached the top voltage and the current is low i reset it.

Recal-high means that if it goes above that value it resets down to 100% again. Same in bottom where it gets back to 0. Thats basically how you do the calibrate or when you have many mons bypass due to high voltage in full charge.

So when doing a top up the mons go into bypass and you eventually go above 100% due to the mons burning off energy. Thats where you have a drift and the system resets to 100% again.

It should not drift and if it do there is something wrongly set in the config i guess. The capacity of the battery bank need to be set based on your cycling scheme ie soc values. So if that doesnt correlate to real life capacity it will not work. Next thing that can perhaps give you issues is if you have losses between the shunt and the battery?


Lets say you fully top off the battery and reset it to 100% manually once. Then you cycle it 5 times and charge back up. You say its then at 90% or something else? Mine is basically spot on 100% i would say. I have been running it for quite a few years and to be honest i dont even look into that any more because its spot on :p
 

gauss163

Member
Joined
Jun 28, 2020
Messages
278
OffGridInTheCity said:
I see (better). One reason Ishy away from SOC in the Batrium context is that it doesn't seem to maintain any accuracy.

I don't have any familiarity with Batrium software, but if you can point my at their docs for your board I will be glad to have a look to see if I can help.
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,581
daromer said:
Lets say you fully top off the battery and reset it to 100% manually once. Then you cycle it 5 times and charge back up. You say its then at 90% or something else? Mine is basically spot on 100% i would say. I have been running it for quite a few years and to be honest i dont even look into that any more because its spot on :p

That's exactly what I have done. Now mind you I have not had my system in operation as long as daromer has but what happens with mine is that if the system goes to 101.5%SOC (Today seems to be a perfect exampleclear sky all day long and great charging) is after a small amount of time it will reset to 100% recalculating SOC.
Here is the example
At 15:54 the system was still charging and showing 101.5% SOC

image_xuciyd.jpg

Then at 15:58 It went into idle mode and reset itself to 100% SOC It may do this 2 or 3 times as the batteries top up to ?4.05V to 4.06V as they are still absorbing some amperage as little as it may be at that point. I am after all not allowing the cells to go to full charge at 4.2V so in theory there is still a lot of amperage these cells could absorb but I am limiting their saturation to ?4.05 to 4.06V.

image_mykmnt.jpg


Wolf
 

daromer

Moderator
Joined
Oct 8, 2016
Messages
5,659
Good example Wolf. Thats the same as here and it just runs. I have synced it so my SOC follows my voltage range i have defined and the available capacity between those.
 

OffGridInTheCity

Active member
Joined
Dec 15, 2018
Messages
1,746
Folks - I just don't get howBatrium settings work on SoC. I rarely get to a hi on any day - its depends on consumption and PV power for the day - but I always hit the low.

Can I use the low - to reset SoC? and how do I do it? For example.... On the Hardware->Shunt page, there is a "Re-Calibrate Low SoC" where I can set a %. Let's say that at 49.5v (3.54v/cell) I believe that SoC should be 20.5%.

If I set "Re-Calibrate Low SoC" to20.5% ... how does this reset anything when the battery reaches the49.5v low? How does the 49.5v come into the algorithm for Batrium to know to do a reset?
 

Wolf

Active member
Joined
Sep 25, 2018
Messages
1,581
Hm I wonder if your nominal capacity entry has some influence.
Also the SoC% cycle threshold.
I have yet to see my system do a SoC re-calibration in the low SoC threshold but then I am not awake at that time. Iwonder if it is in the log file somewhere.
I would imagine it is.
Here is a screen shot of my settings.
Wolf


image_bmxcwa.jpg
 

gauss163

Member
Joined
Jun 28, 2020
Messages
278
daromer said:
Lets say you fully top off the battery and reset it to 100% manually once. Then you cycle it 5 times and charge back up. You say its then at 90% or something else? Mine is basically spot on 100% i would say. I have been running it for quite a few years and to be honest i dont even look into that any more because its spot on :p

How does it keep FCC = full charge capacity updated if one never does full cycles?

I tried to find documentation on how they define/calculate SOC on their website via google searches and came up empty. Is the documentation really that poor? Surprising if so given its high price.
 
Top